LOAD BALANCING DATA-PARALLEL PROGRAMS ON DISTRIBUTED-MEMORY COMPUTERS

In this paper a set of programming constructs for the implementation o f data parallel algorithms on distributed memory parallel computers is proposed. The load balancing problem for data parallel programs is ca st in a special from. Its relation to the general load balancing probl em is analyzed. The applicability of these constructs is asserted for a number of grid-oriented numerical applications. A software tool prov ides run-time support for data parallel programs based on the proposed constructs. While the application - according to the data parallel pr ogramming paradigm - partitions the grid, the tool assigns the partiti ons to the processors, using built-in mapping algorithms. The approach is general enough to accommodate for data parallel algorithms with va rying communication structure and variable calculation requirements us ing pseudo-dynamic load balancing strategies.